Adds some punctuation massaging.

hzafar · Jan 5, 2021 · b3b0963 · b3b0963
1 parent 43b8a14
commit b3b0963
Show file tree

Hide file tree

Showing 3 changed files with 58 additions and 7 deletions.
diff --git a/src/main/kotlin/ca/voidstarzero/isbd/ParserUtils.kt b/src/main/kotlin/ca/voidstarzero/isbd/ParserUtils.kt
@@ -55,6 +55,27 @@ fun droppedPeriods(input: String): List<String> {
     }
 }
 
+// idea is we want to detect improper punctuation and
+// try to correct it to give a better parse. this may
+// ultimately not be the right way to do this.
+fun punctuationFixes(input: MARCField): MARCField {
+    // if colon at subfield boundary is missing preceding space, add it in
+    val withColonSpaces = input.subfields.map {
+        if (it.second.matches(".*[^ ]:$".toRegex())) {
+            val fieldData =  it.second.substring(0, it.second.length - 1) + " :"
+            Pair(it.first, fieldData)
+        } else {
+            it
+        }
+    }
+
+    return MARCField(
+        input.tag,
+        input.indicators,
+        withColonSpaces
+    )
+}
+
 fun usesISBD(input: MARCField): Boolean {
     if (input.subfields.any { it.first == 'c' }
         && !input.fieldData().contains(" / ")) {

diff --git a/src/main/kotlin/ca/voidstarzero/isbd/titlestatement/TitleStatementParser.kt b/src/main/kotlin/ca/voidstarzero/isbd/titlestatement/TitleStatementParser.kt
@@ -1,13 +1,10 @@
 package ca.voidstarzero.isbd.titlestatement
 
-import ca.voidstarzero.isbd.cleanInput
-import ca.voidstarzero.isbd.prepare
-import ca.voidstarzero.isbd.simpleParse
+import ca.voidstarzero.isbd.*
 import ca.voidstarzero.isbd.titlestatement.ast.TitleStatement
 import ca.voidstarzero.isbd.titlestatement.grammar.TitleStatementGrammar
 import ca.voidstarzero.isbd.titlestatement.grammar.monographRoot
 import ca.voidstarzero.isbd.titlestatement.grammar.seriesRootWithMARC
-import ca.voidstarzero.isbd.usesISBD
 import ca.voidstarzero.marc.MARCField
 import ca.voidstarzero.marc.fieldData
 import norswap.autumn.Autumn
@@ -59,7 +56,8 @@ class TitleStatementParser : TitleStatementGrammar() {
         }
 
         val parseRoot: rule = seriesRootWithMARC(marc)
-        val result = Autumn.parse(parseRoot, cleanInput(marc.fieldData()), ParseOptions.get())
+        val preprocessedMarc = punctuationFixes(marc)
+        val result = Autumn.parse(parseRoot, cleanInput(preprocessedMarc.fieldData()), ParseOptions.get())
         if (result.full_match) {
             return result.value_stack.mapNotNull { it as TitleStatement }
         }

diff --git a/src/test/kotlin/ca/voidstarzero/isbd/titlestatement/SeriesTitleTest.kt b/src/test/kotlin/ca/voidstarzero/isbd/titlestatement/SeriesTitleTest.kt
@@ -38,8 +38,10 @@ class SeriesTitleTest {
                             )
                         ),
                         entrySors = listOf(
-                            SOR("Bulgarian Academy of Sciences, Centre for Scientific " +
-                                    "Information and Documentation")
+                            SOR(
+                                "Bulgarian Academy of Sciences, Centre for Scientific " +
+                                        "Information and Documentation"
+                            )
                         )
                     )
                 )
@@ -1041,4 +1043,34 @@ class SeriesTitleTest {
         assertNotNull(result)
         assertEquals(expected, result)
     }
+
+    @Test
+    fun t28() {
+        val marc = MARCField(
+            "245",
+            "|aState summary:|pType and amount of aids paid to all local governmental units and counties.",
+            '|'
+        )
+
+        val expected = listOf(
+            TitleStatement(
+                titles = listOf(
+                    Series(
+                        seriesTitle = SeriesTitle(
+                            title = "State summary",
+                            otherInfo = listOf(
+                                SeriesOtherInfo("Type and amount of aids paid to all local governmental units and counties")
+                            )
+                        )
+                    )
+                )
+            )
+        )
+
+        val result = t.parseSerial(marc)
+
+        assertNotNull(result)
+        assertEquals(expected, result)
+    }
+
 }